by Michael D. Anestis, M.S.
In a recent post on the use of empirically supported treatments (ESTs) for eating disorders, I spent some time discussing various controversies surrounding the use of ESTs. One issue, however, that I did not dwell on nearly enough was the idea of using group data to determine the utility of a form of therapy even though all of us are unique individuals. I am thankful to Cara, a reader who chose to comment on that article, for calling this point to my attention and today I would like to make a more formal attempt to clarify the issue.
Over the past several years, I have encountered many individuals who have passionately argued that group data from studies are not useful in determining which form of therapy is the most useful for a particular individual. In my critique of this point, I want to be clear in emphasizing that neither I nor any other proponent of scientific research believes that all clients are the same. We recognize very clearly that each client brings a unique perspective, life story, and set of goals to the therapy room. Additionally, few if any believe that any single therapy will work for everyone, regardless of how strongly supported that approach might be. Instead, scientists maintain the belief that treatments should be systematically studied and compared to one another with respect to how well they address the symptoms of a particular mental illness and to what degree they prevent those symptoms from returning in the future. Based on these results, therapists should then prioritize using the approach to treatment that has been shown to be the most effective. The rationale for this is that, by using the approach that typically produces the best results, the clinician will maximize the odds that he or she will successfully treat their client on the first try.
This next point is often a point of confusion. Even the most well supported treatments will, on occasion, not work. This could be for a variety of reasons. The therapist might not properly administer the treatment, the client might not complete homework assignments, or some unknown variable might simply make that particular treatment not a good match for that particular client. Perhaps a combination of several factors contributed to treatment failure. If this situation arises, empirically minded psychologists believe the next step is to try alternative ESTs. If none are available, then treatments without empirical support might become options (although, personally, I would favor referring the client to another clinician who could try the EST in a different environment in case the particular therapist played a role in the initial treatment's lack of success). So, the argument here is not that ESTs always work or that alternative treatments never work. The point is that ESTs appear to work more often and, as such, should be prioritized as the front line approach.
This brings us back to central question though. Sure, these studies show what happens on average, but how does that tell a clinician whether or not the favored treatment will work for a particular client? In short, it doesn't; however, nothing else does either. On this point, I think scientifically minded and non-scientifically minded psychologists actually agree more than they think they do. If there was a way to determine who will and will not respond well to a particular treatment, everyone would favor using that approach. In other words, if we could identify ahead of time the clients that, for one reason or another, might benefit from psychodynamic therapy for their depression instead of cognitive behavioral therapy, CBASP, or interpersonal psychotherapy, that would be fantastic. In fact, there is a type of statistical procedure called moderation analysis that can help reveal whether a specific variable impacts the nature of a relationship. So, there actually are ways to determine if, for example, a particular demographic group is more likely to respond to a specific form of treatment. Thus far, however, we do not have data to guide us on that front and the clinicians who choose not to use ESTs are not making that decision based upon analyses of moderators. Instead, they are basing it off of intuition. This is where I see a problem and a good teaching point to explain why group data is useful.
Without question, some individuals will not significantly improve even through the best form of treatment. If we cannot determine ahead of time who will and will not benefit from treatment, that leaves us with nothing but guesswork. By using an EST, we guarantee that folks who simply were never going to respond to that treatment will not see a quick reduction in their symptoms. The same can be said in reverse, however: if a client was not going to respond to a non-empirically supported or eclectic form of therapy, they were in a bad spot in the beginning as well if such an approach was taken. So when we use ESTs, we are accepting that some individuals will likely need to try a different approach before their symptoms improve, but we accept this cost because, in doing so, we will actually successfully treat the highest possible percentage of our clients. By guessing, we might win big and correctly pick the best treatment for each specific client, but what are the chances of that? Let's say that 80% of clients respond well to treatment A and 30% of clients respond well to treatment B. Clearly, this means some clients will respond equally well to both treatments. It also means that some individuals might not respond well to either option. If we treat our client with treatment A, there is a 20% chance that he or she will not respond well to this initial approach. If we begin with treatment B, there is a 70% chance that he or she will not respond. Why then, would we not start with treatment A and then switch to treatment B if improvements are not made during the course of the first several sessions? Certainly, treatment A will not work for everyone, but if we simply make guesses, that number is almost certain to creep well beyond 20%. Obviously these numbers are arbitrary - I made them up to emphasize a point - but they serve to illustrate the issue nonetheless. Yes, studies look at groups and groups do not tell us about specific individuals outside of that sample; however, without any other source of information regarding a specific individual and his or her likelihood of responding well to a particular treatment, group data affords us our best opportunity to provide the best possible care to the greatest number of clients. Some clients will require a shift from the initial treatment to an alternative but, you know what....that's true for folks who do not use ESTs as well.
Many clinicians will argue that, through experience, they have developed an eye for how to simply integrate components of many treatments together into a personalized approach for each individual that enters their office. When pressed on how they determine their success, many will point to positive reviews from their clientele. Isn't this simply another form of group data though? Even more, isn't it a less reliable form of group data than what is found in published studies that make data available to all readers and thus leaves findings capable of being examined by independent sources? Alternatively, a clinician might say that females or Hispanic individuals or individuals above a certain age are more likely to respond to an alternative treatment. This, however, is an easily tested hypothesis and, without data to support those assumptions, they represent risky gambles, not informed choices. If a clinician feels strongly about this, they should measure treatment gains in their practices and publish the results, as such information would be invaluable for the entire field.
This brings us to the take home point. Group data does, in fact, reflect what happens on average. When scientists use group data to drive treatment decisions, however, this does not mean that they are assuming that you, the client, are average. Instead, they are saying that, in general, one particular therapy tends to lead to the best outcome. As a result, they prefer to begin therapy with that approach, as it offers the greatest likelihood of success. At the same time, they realize that the treatment, like anything else, might not work and, in that scenario, a later change might be necessary. Because symptoms of mental illness are often debilitating and, in some cases, mortally dangerous due to an increased vulnerability to suicide, it makes sense to prioritize the use of treatments with the most research support, as this approach is likely to lead to a quick response the most often.
As I close out this article, consider a parallel to medical interventions. Let's assume there is a surgical procedure that has been shown to be highly successful in treating a deadly disease. In addition, there is an experimental procedure that has not been studied as regularly and has not produced consistent results, but for some individuals has been shown to work quite well. In light of the sporadic promising results, the alternative treatment is alluring, although there is no data indicating who are the lucky ones who will respond well to its methods. It is entirely possible that the alternative approach will be what works best for you, but how could you know? You could guess and you could be right, but the same theory is what drives people to play the lottery. Mental health treatment deserves the same type of consideration as treatment for medical ailments. In order for that to happen, however, we need to be guided by group data.
I fully recognize that many of you will likely disagree with my points - some of you vehemently - and I respect that. What I would love would be for folks on both sides (or from alternative perspectives) to voice their opinions and to open a dialogue on the matter. The key for me is that, whether or not you believe in ESTs, you know that researchers are not under the impression that ESTs work for everyone or that the group average reflects what each individual client will feel. Instead, they believe that we produce better overall results by sticking to what has been shown to work and that, as new treatments emerge, progress will continue to be made. They are not simply campaigning for CBT or any other particular treatment in order to protect their professional positions. They are campaigning for scientific skepticism that forces us to test whether what we feel reflects what actually happens or whether clients would benefit from an alternative approach. I look forward to hearing your thoughts as always.
If you would like to read more about empirically supported treatments, please consult the many resources available through our online store.
Mike Anestis is a doctoral candidate in the clinical psychology department at Florida State University
Subscribe in a reader
